Académique Documents
Professionnel Documents
Culture Documents
. .
otlmxzation and Characteristics of a Sailwing
Findmill Rotor
by: M.D. Maughmer
Published by:
Research Applied to Nalionai Reeds
National Science Foundation
Washington, DC 20550 USA
Paper copies are $8.00 Ask for accession number
PB-259 898 when ordering.
Available from:
National Technical Information Service
Springfield, VA 22151 USA
PB-259 898
Prepored for
Mar 76
MEET
,. Title and Subtitle 5. Report Date
lptimization and Characteristics of a Sailwing Windmill Rotor March 1976
IFinal Report, February 1, 'l975-January 31, 1976) 6.
r. Author(s)
I.D. Maughmer
1. Performing Organization Name and Address 10. Project/Task/Work Unit No.
'rinceton University
Department of Aerospace and Mechanical Science 11. Contract/Grant No.
'rinceton, New Jersey 08540 GI 41891
12. sponsoringOrganization Name and Address 13. Type of Report 61 Period
. Research Applied to National Needs (RANN) Corered Final Report,
ilational Science Foundation 2/l/75 - l/31/76
Jashfngton, D.C. 20550 14.
l’ I1
15. Supplementary Notes
16. AbscracrsWithin this fourth and final quarter progress report are comprehensively dis-
:ussed all of the research efforts undertaken by the Princeton windmill group over the
last year. This includes a detailed accounting of the development and operational tech-
iiques of the Princeton moving-vehicle windmill testing facility. Also presented is a
:omplete documentation of the performance build-up (DPmax = .06 to Cpmax = .4u) of a 12
Ft. diameter, two-bladed Sailwing rotor. This report further includes an examination of
In exploratory research effort directed toward using a small, first-stage, co-axial
rotor to augment windmill performance. Finally considered are the results and conclu-
sions of an extensive wind-tunnel test program aimed at a quantitative determination of
the aerodynamic penalt.ies associated with numerous simplifications of the basic double-
nembraned Sailwing cross-section.
18. Availability Statement 19. Security Class (This 21. No. of Pages
NTIS Report)
UNr-I.ASm
20. S_curicy Class (This
Page
FORM NTIS35 IREV. IO-,I)
I UNCLASEIFIED
ENDORSED BY ANSI AND UNESCO. THIS FORM MAY BE P.EPRODUCED lJdCO”M-DC 02es-p74
OPTIMIZATION AM! CHARACTERISTICS
Mark D. Maughmer
March, 1976
by
Mark D. Maughmer
Princeton University
08540
March, 1976
Final Report
Approved by:
Thomas E. Sweeney
Report : NSF/RANN/GI-4189l/FR/75/4
Within this fourth and final quarter progress report are compre-
Princeton windmill group over the past year. This includes a detailed
CONCLuI>mG SuMMARy - m - - - - - - - - - - - - - - - - - - - - - - 28
29
~w~~I"""""""""""""-""""""""""""~" 30
APPElqI)MII -------------------------- - 31
. FIGURES
ii
LIST OF SYMBOLS
b Wing span
S Wing area
iii
INTRODUCTION
of the Princeton windmill research group for the past year. From all
3
available information, in the case of the smaller-;ized family of wind
machines for which the Princeton group has been, and remains, a strong
performance with its rigid-bladed counterparts and yet enjoys the benefits
Sailwing windmill in hand, further pursuance of our goal, which has been
the effort. While most certainly the moving-vehicle test facility will
build-up technique that has been employed by the Princeton windmill group
great deal of effort required to obtain those last few percentage points
that have been obtained up until now were "put to good use" and the
feeling is that the time is now right to proceed toward the practical
1
Y
that the Princeton group would like to direct the expertise it has
Y
THE PRINCETON JEEP WINDMILL TESTING FACILITY
and rapid development of rotor performance that was the goal of the
dramatic effect on the net power generation when integrated over a long
begun. The gradual evolution of this facility has now reached a point
experience. Due to both the reasonably low scatter and the high repeat-
ability of the data collected using the jeep windmill tes;ing facility
3
-
-- -~I_.-
--
anemometry systems were tried before the successful type shown in the
orerall views of the test vehicle, Figures l-3, was employed. These
attempts included both mechanical vane and hot-wire types until the
c
driver to match the speed of the vehicle to the wind velocity desired,
Both the mechanical and the hot-wire types proved to be too susceptible
L
ing, yet, due to its light-weight, halved ping-pong ball construction,
verified before each series of tests by recording the time required for
the jeep to travel a given distance and comparing the resulting calcu-
4
the additional purpose of indicating whether or not the natural winds
several methods of loading the windmill were tried, the last and most
The master cylinder and actuating mechanism for this system are mounted
sitting in the bed of the Jeep. 'I.%; brake is applied by turning down the
ment signal to th2 dctua:.or-arm with a :. ..' ,ant fc.,:,ce signal. The disk-
brake itself is ahmn against the larger r~;;lus rotor hub assembly in
Figure 5. The amount of torque that is gs verated by the rotor when the
aiIL-tint, The brskiilg system is calibrated by locking the disk and noting
of a small tachometer mounted at the rotor hub, the signal from which is
sent to a voltmeter, Figure 7, and read by the crew member riding in the
5
right seat of the Jeep. Calibration of this unit is performed by rotating
the hub by means of an electric drill and correlating a strobe light RPM
B) Data Collection
the Jeep windmill testing facility is that of very light or, preferably,
sample. This requirement for absolute calm test conditions is only met
with any regularity in the early dawn, and, thus, most of the Princeton
data has been oqtained during the couple of hours between first daylight
and sunrise.
for each pass down the 3000 ft. Princeton University runway. The man in
be applied, reading the RPM meter, and for recording all of the informa-
ining when all of the readings have stabilized sufficiently for a point
to be taken. Finally, the man sitting in the bed of the Jeep maintains
the torque setting required for a given point by watching the torque-
6
testing vehicle become experienced at their posts and familar to the
C) Reduction of Data
tion and compare its operation to that of other designs. For optimiza-
tion research, such as that performed by the Princeton group, this capa-
the tests should stiE be ‘valid in that the data is reduced on the basis
of the indicated airspeed of the vehicle and should be the same velocity
c was found that because of a vertical wind velocity gradient above the
test strip (boundary layer), along with the fact that the windmill and
the anemometer were not mounted at the same elevation above the ground,
such was not the case. Thus, in an effort to extend somewhat the conditions
7
under which useful data could be obtained, a method of correction was
developed for hose occasions when a light wind was blowing parallel,
Appendix 11; however, the result of its use can be seen by comparing
Figures 9 and 10. In Figure 9, essentially two separate plots have been
heavily toward the higher curve in Figure 9. All in all, the best data
has added a little flexibility to the Princeton test program and has
formance and which elements can be eliminated with the benefit of lower-
configuration in both pitch and twist. Thus, it was hoped that such
two-bladed rotor was chosen for the basic design because of the much
ever, even though such problems are m.inimal because of the low inertia
does recognize the advantages of and strongly recommends the use of three-
b
9
-. .~ ---_._
The basic Sailwing structure from which all the subsequent con-
(manualiy rotatabiej and tip members. Tie blade trailing edge consists
trailing edge produces chordwise tension in the sail itself which can,
center-body disks (d) and (g), and various types of trai.ling edge stiff-
angle was adjusted until the performance was found to drop-off on either
stream wind velocities of 19.6 ft/sec (13.3 mph) and 31.3 ft/sec (21.3 mph).
The data for each configuration and wind speed that was evaluated
Figure 13. It will, of course, be noted that the msximum power coefficient
edge should be documented to answer one of the most often posed questions
blade tip cross-section of this configuration which was the only one with
D-tube fairing to the leading-edge, the power coefficient has been in-
11
relatively large thickness ratio of each saction along the blade span.
Also shown in Figure 14 is the blade tip profile that was used on all of
the tipless version, it was aore than balanced by the increased rotor
efficient. The fact that the power coefficir:nt itself did not improve is
ing a radius 20% that of the overall blade radius, Figure 12(d), resulted
and 19.
This brace between the leading-edge spar and the trailing-edge cable
at that point but does not effect the basic catenary arc of the trailing-
12
by the configuration shown in Figure 12(f) but resulted in a little
power coefficient that was obtained, Cp = .40, for the iower wind speed,
Figure 25. The higher wind speed performance is presented in Figure 26.
.
(4 4.0 .06 2.8
13
B) Conclusions
further be stated that the addition of ging tip fairings on this configura-
tion did not improve the overall power coefficient obtained. The instal-
lation of a 20% center-body disk did modestly improve the rotor performance
basic machine improves, the optimum blade root pitch angle decreases from
twenty degrees to ten degrees as the tip-speed ratio at which the maximum
power coefficient occurs increases from 2.8 to 4.3, Somewhat less ex-
pected is that fact that the twist distribution is not particularly criti-
degrees of optimum. This is perhaps due to the fact that in the case of
relationship between the root and tip ribs. Thus, as the mid-span sections
the actual positioning of the tip relative to the root is lass important
the wind velocity from 19.6 ft/sec to 31.3 ft/sec resulte:d in an average
14
deterioration of approximately eight percent in the wximum power co-
crucial, some attention must be given to the fact that this condition
will become more pronounced as the velocity approaches that of the criti-
sail luffing occurs (Reference 1); however, lest the impression is given
that this efi'ect must be entirely detrimental, it has been suggested that
tion with increasing wind speed can be used to aid in the prevention of
wing rotor build-up with those of the three-bladed rotor, Figurs 1 and
2, it was expected that the lower overall solidity ratio of the two-
needed to obtain .44 (Reference 1). While the explanatLcn of this dilem-
overall more optimum induced velocity distribution over the disk area,
this has yet to be determined rigorously. Thus, even though both rotors
are considered excellent when compared to other modern machines, one can
appreciate the fact that there remains a great deal to be learned con-
15
\- MULTI-STAGE WINEMILL TO @TAIN C”,‘mR-BODY FAIRING BENEFITS
performance that was realized with the addition of a 20% centere.body disk
rotor, a multi-stage rotor assembly was built and tested, Figures 27 and
stream rotor were constructed in such a manner that the pitch could be
the course of testing to apply a variable load to the first stage rotor
from that of the main rotor. This capability was included because it
was not intiutively clear whether the flow field would be most beneficial-
the main rotor in order to put energy into the center portions of the
16
the outboard regions. Obviously, a very broad exploratory program wculd
be required to determine th- most suitable operating mode and only then
concept evaluated.
17
B) Conclusions
In examining the data presented in Table II, one can only come
yielded performances that were either the same or slightly worse than
that of the baseline windmill. In any case, the upstream rotor’s effect
that while the upstream rotor may very wall produce a fluid-flow dis-
turbance much like the center-body disk, it also produces and “in-plane”
interference due to the drag of its blades. Thus, any beneficial inter-
ference or additional rotor torque is nullified and the net rotor per-
conclude that a simple center-body disk is not only more effective, but
18
WIND TUNNEL fNVESTIGATION OF VARIOUS SAILWINC AND SAILVANE CROSS-SECTIONAL
PROFILES
was tested by including several sections not utilizing the lower membrane
wings, identical in all respects except for the section utilizej, were
tested.
A) Model Description
The wing planform that was utilized is shown in Figure 36 and character-
4.5 inches, and a total area, S, of 168 sq. inches. The resulting aspect
2
ratio of the planform, AR = b lb; , is equal to 8.4. Relative to the
and was adjusted to 9.5 ibs and 36 lbs for the experimental results
19
discussed herein. Respectively, these values yield a trailing-edge
in this report were obtained with the tunnel speed adjusted to yield a
and +24 degrees. Thus, force balance data for lift, drag, and pitching
moment were obtained at each two-degree angle increment between the limits.
These data were then reduced to the standard coefficient form and plotted
pointed out is concerned with the fact that all of the drag measurements
are obtained by subtracting a very large tare drag reading from a very
large overall drag reading to obtain the very small drag contribution of
the model. Although this is the common procedure when a limited amount
20
of time and funding are available, it undoubtedly allows small errors in
the drag measurement to enter into the experiment. This is not to say
that the procedure yields data which is unsuitable for design purposes
when the Sailwing is at rest (no-wind), the cloth membrane is held taut
a symmetrical section as the upper and lower surfaces experience the same
lished between the upper and lower surfaces causes the membrane (or
membranes) to deform away from the high pressure regions (underside) and
toward the low pressure regions (uppdrside), Thus, when the wing is
lifting upward, Figure 38, the airfoil section assumes a positive camber
distribution that fairs in smoothly with the airfoil’s leading edge shape.
of lift that is generated (up until the wing stalls), the resulting
21
increased pressure differential between the upper and lower surfaces
causes the amount of camber in the section to increase. This not only
laxing the cable tension allows for a greater mount of camber and there-
it might be considered that as the cable tension becomes higher and higher,
the Sailwing's behavior becomes more and more like that of a rigid wing.
reduced angle of attack in these regions and one would expect a local
more than offset by the increased amount of camber that occurs and results
the lift distributions that occur over some Sailwings are often very
edge cable and the catenary arc sail cut is responsibie for many of the
22
desirable features of the Sailwing over other flexible designs. One
such feature is that realtively low drags are present at low angles of
attack and lfft coefficients. Furthermore, the Sallwing has the ability
Figure 41.
to the conventional rigid wing while not suffering any performance penal-
tests of the eight Sailwing and Sailvane models are presented in the
plots of Figures 42-57. For each wing Lested, the lift coefficient, drag
have been obtained from these data are summarized in Table III.
23
the cast: of a Sailwing to linearize the drag polar or obtain a meaning-
Y
angles-of-attack (up to approximately five degrees), most Sailwings have
a lift-curve slope which exceeds the theoretical maximum for rigid wings
(27t per radian = .ll per degree). This occurs because the section is
values.
Section dCL/
dor (per degree) ~Lo(dewedQ.,~x (I/D)mFz (L/~)cI=l.o
concluded that for the dynamic pressure at which the tests were performed,
both of these values are sufficiently high to allow only minimal increased
al data from three-dimensional data such as that presented for the Sail-
wing; however, the flexibility of the Sailwing airfoil makes this extremely
difficult and the Princeton group does not know of any reasonable method
would require that the drag polar be linearized to some degree such that
rigid wing utilizing a similar profile, for example the NACA 4412, it can
that of the rigid section to a large extent and it is not at all unreason-
such that the maximum sectional lift-to-drag ratios are of the same
25
indication that the sectional efficiency of the Sailwing does not differ
E) Conclusions
evident that the performance penalties paid for deviating from the con-
the leading-edge shape are th% larger values of lift coefficient, along
leading-edge gives rise to a slight shift in the entire lift curve such
is quite competitive with most rigid wings of the same aspect ratio.
Thus, the use of the Sailwing should allow the benefits of simpler con-
the fact that, unlike many of its rigid counterparts, the cambering
26
to-drag ratios to be very near maximum at lift coefficients of close
27
1. The Princeton moving-vehicle windmill testing has reached a level of
bladad Seilwing rotor resulted in the value of the maximum power co-
blade technology.
29
APPENDIX I
Contributors
30
APPENDIX II
.
It has been found by the Princeton group that when the moving-
vehicle test facility is used to collect data in the slightly less than
31
.
senses two different true windspeeds; however, the free-wheeling tip-
condition
sL,R
‘1 true
R, = vindicated + r
nz. ‘indicated - *v”
Solving for 1p yields,
w-= (Vindicated) 3 -
\)/(
L 3
Thus, if the ratio of the upwind to downwind free-wheeling angular
‘U and, therefore, the true wind velocity for each direction can be
reducing the data collected, the true performance curve of the windmill
that the wind has been reasonably constand and that the correction is
valid.
32
PRINCETOX JEEP WINIMILL TESTING FACILITY
FIGURE 2
33
W
P
, .
FIGURE 4
FIGURE 5
DISK-BRAKE ASSEMBLY
35
.
FIGURE 6
TORQUEREAD-OUT
FIGURE 7
ROTOR TACHaMETERREAD-OUT
36
46 0706
1 , . 10 X 10 TO THE INCH.
KEUFFP L ESSER co
7 X IO lN~11t.a
1.01 I9 Y I1
(
46 0706
c
.-
-_
t-
_-
__-- -
. _-
”I’:‘I
I
f
FIGUKE12
!
v----c -_----------_---------------
-“’ ---...---------- TT -------------- a-
-i- r PULL a
I 7
wwn ne mnit4~
1 :
:
I
I
.
TL
<
Y
.
\ .
( ( t .
tw IO X 10 TO ME
KEuFFn L UiMn
INCH. 7 X IO INCHES
CO. mm I” . . . 46 0701,.
c
C ( ( ( !
I44 IO X 10 TO THE
KEuFmaEsst34co.
INCH. 7 x to INCNES
“*o(I”oIL 46 0706
C
c 1 -.-- -‘----T- ‘- - ,..,-0
46 070:. ,
I
C
46 07d i
I 1 .
. 10 X IO TO THE INCH. 7 X 10 :NCHES
KEUFFP. 8 E!SSER CO ua II Y s.
c ( c (
t+E 10 X IO TO THE
XElJFFa a Essm
INCH. 7 x IO ,NCHES
co. uaotYI(SA 46 0706
C
I - I ( ( f ( ( (
WE 10 X 10 TO THE
KKUFFP. L EssfR
INCH. 7 X10
co. “101. * ILL
INCHES
46 0706’ ’
( ( ( (
WE 10 X IO TO THE INCH. 7 x IO INCHES
NEUFFEI. * ESSER co. rra # Y&L 46 0706
c
( ( ( t ( ( t (
HOE IO X IO TO THE
KEUFFR
INCH.7 X IO INCHES
a ESSER co. “‘DE I, LL 46 0706’ ’
I c t ( ( ( ( f t (
He IO X i0 TO
UEUF?EL
THE
& &SW
INCH
CO
s 7 x 10 WLHES
Y.D( m I L. 46 0706
c
\ ‘8 I- ’ IO X 10 TO
KW~LESSERCO.
THE INCH. 7 X 10 INCHES
IIDII.LL 46 0706’
4 t #ai 4 (
IO X 10 TO THE INCH.
4 I
7 X 10 INCHES
s f
KQJFFEL~cssERco. UalYLL
c
46 0706, ,
CC
Y
FIGURE 27
WINDMILL ROTORASSEMBLY
FIGURE 28
-_I
CLOSE-UP OF THE CO-AXIAL WINDMILL ROTORASSEMBLY
46 0706 ’
1 (
46 0706
6
I f
46 0706
4. *
H*E IO i IO-TO THE
KEUFmhESSERCO.
INCH. 7 X 10 INCHES
MWKIUS~ 46 0706
C
t t t !
46 0706
SAILVANE MODEL 1
SAILVANE MODEL 2
0 SAILVANE MODEL 3 -
a/ SQlI-SAILWING
- -------
MODEL 3
SAILWING MODEL 2
ii
FIGURE 37
(NO WIND)
FIGbRE 38
LOADED SAILVANI?
65
FIGIN 39
--SAILVANE POSlTIVELY
CAPSEUD IN A LIFTING
CONDITION
--FIGURE 40
PHOTO SHOWINGTRAILING
- EPCE DEFORMATIONS
!!XER LOAD
FIGURE 41
SAILVANE IN A NEGATIVE
LIFTING CONDITION
c
. SAILVAIiE MODEi-
68
FIGURE 44
SAILV~WE NODEL 2
SAILVANE KODEL 2
FIGURE 45
70
c
t
Y
Y
------- -
SEMI-SAILW'liNG MODEL 1
SEMI-SAILWING MODEL 2 FIGURE .F
T
i-.-
(-J./ 1
SEMI-SAIT,WING MODEL 3
-__
SAILWING MODEL 1 FIGURE 54
79
Y
Y
80
I
‘.
/
SAILWING MODEL 2